[lacnog] Bogon route objects in the LACNIC IRR
Ronald F. Guilmette
rfg en tristatelogic.com
Sab Ago 14 07:15:27 -03 2021
Job Snijders suggests that DFZ routing of IP address space using
unassigned "bogon" AS numbers is somehow necessary. In support of
this assertion he provides only further unproven assertions, which
he himself has made, and which he asks us all to blindly believe:
"For sound operational reasons both in the RIPE database and in the RPKI
in general, the IP resource holder is permitted to designate any ASN
they wish as the origin."
(Please note that Job does not actually bother to list any of these supposed
"sound operational reasons".)
"In many RPKI deployment scenarios it is *** technically impossible ***
for the RIR to impose restrictions on the content of the ASId field,
because the RIR is not involved in the issuance or publication of said
Well, it would seem that *I* am able to achieve what has previously been
deemed by experts to be "technically impossible". Maybe I should rush down
to my local patent office and file a patent before anyone can beat me to it!
With all due respect to my friend Job, the above statement is apparently
equivalent to saying "Ron Guilmette, working all by himself and with -zero-
funding, can perform spectacular feats of magic that even the five Regional
Internet Registries, with all of their money and with all of their highly
trained and experienced staffs cannot do --- He can see, on any given day,
which AS numbers are unassigned "bogon" ASNs and which ones aren't.
Job has a perfectly valid point that no individual RIR can ever be absolutely
and positively 100% sure that a given *assigned* ASN, particularly one that
was issued and assigned by some -different- RIR, is or is not a valid thing
to be using as the origin ASN for any given route to any given IP address
block. But we are *not* taking about "valid and assigned ASNs" here.
That is not the issue, and that is not the question. That is just an
The issue isn't whether or not some valid and properly registered ASN
may or may not be used to route a given IP address block. The issue is:
"Do we want people making route announcements into the DFZ while using
provably INVALID (bogon) AS numbers?"
I believe that anwer must be "no" and that thus it follows that no RIR
should be, in effect, endorsing the use of bogon ASNs, e.g. by allowing
route objects that refer to bogon ASNs into their own IRRs.
And it clearly does not require any wizard level special magic in order
to see, on any given day, which ASNs -have- been properly assigned by
some RIR to some resource member and which ones haven't (and are thus
"bogons"). These distinctions may be trivially made by anyone with even
the minimal skill necessary to (a) download the daily "stats" files from
the five different RIRs and then (b) apply to those some very minimal
textual manipulation. (This really is not rocket science, and I do believe
that even Job could do it, if he put his mind to it.)
Fortunately for the various RIRs, I detest bogons and thus, I have already
done all of the work necessary to find and report these bogon route objects...
objects that refer *either* to invalid/unassigned IP addresses *or* to
invalid and unassigned ASNs.
Some RIRs, at least, have thanked me for my efforts and have already removed
the bogon route objects in question. (LACNIC isn't one of them.)
Job is quite clearly and vigorously arguing in favor of the proposition
that the Internet should be, in effect, a lawless free-for-all where any
party may use *any* arbitrary ASN... even provably bogus and unassigned
ones... within route announcements that they then inject into the DFZ,
*and* that the five RIRs should not only tolerate this aberrant and anti-
social behavior, but that they should even endorse it by including bogon
route objects into their own IRRs.
As noted above, there is no rational basis for such a choice, or for such
an attitude, and Job's reference to RPKI is an irrelevant smokescreen.
The needs and limitations of RPKI are entirely irrelevant to the ongoing
and parallel operation of (non-cryptographic) IRRs by the five Regional
Internet Registries. These two realms are, as we say here, "apples and
oranges", and there is no need to copy the mistakes of one realm onto the
other or vise versa.
That having been said, when and if RPKI does eventually become more widely
deployed it is my hope that *someone* will be looking at all of *those*
RPKI-secured route objects from time to time, to see if any of *those*
things mention any bogon ASNs. If any do, then someone will need to clean
up that mess also.
As we say here "It's a dirty job but someone has to do it." Otherwise the
DFZ will become like an untended garden... full of weeds and not much else.
Job is, I believe, simply confused. He believes that I wish to uproot some
of his prized tomatoes from the garden. But I don't. I'm only interested
in getting rid of the actual weeds. And these actually are not at all
difficult to differentiate from the stuff that Job wants to keep. Indeed,
I have already done exactly that.
P.S. One of the more humorous aspects of this debate is that actually,
as far as I can tell, exactly -zero- of the 20 bogon route objects that
I previously listed here are even being used, at present, to route anything.
These are the routing equivalent of the proverbial "Tree that falls in the
forest when nobody is around."
Perhaps Job could also be the press spokesman for Sri Lanka's Mattala
Rajapaksa International Airport, since that also appears to have no
actual reason for existing:
Más información sobre la lista de distribución LACNOG